Do you qualify?
Machine learning and AI engineering is a shortage occupation in Germany under ISCO-08 group 25 (ICT professionals). This means you qualify for the EU Blue Card at the lower salary threshold and the IT specialist exception applies if you have experience but no formal degree.
To qualify, you need all three:
- A job offer in Germany with a contract of at least 6 months
- A gross annual salary of at least €45,934.20 (2026 shortage threshold)
- A recognised university degree or at least 3 years of ML/AI experience in the last 7 years
Does "ML Engineer" qualify as a shortage occupation?
Yes. The Federal Employment Agency classifies machine learning engineers, AI engineers, data scientists, MLOps engineers, and research engineers under ISCO-08 group 25 — typically unit 2519 (software and applications developers NEC) or unit 2521 (database designers and administrators), depending on whether your role is model development or data platform oriented.
Non-standard titles — "AI Engineer", "Applied Scientist", "Research Engineer", "MLOps Engineer" — are classified by actual work content, not job title. Any role involving the design, training, and deployment of ML systems at graduate level qualifies under group 25. This is confirmed on your employer's Erklärung zum Beschäftigungsverhältnis form.
Salary threshold for ML/AI roles (2026)
| Category | 2026 minimum gross salary |
|---|---|
| ML / AI engineer (shortage occupation — ISCO-08 group 25) | €45,934.20 / year |
| Any profession (general threshold) | €50,700 / year |
| IT specialist without a degree (§ 18g(2)) | €45,934.20 / year |
Mid-level ML engineering roles in Germany in 2026 typically pay €60,000–€90,000. Most offers clear the threshold comfortably.
Two routes to the Blue Card
Route 1: University degree
Your degree must be comparable to at least a German bachelor's (ISCED 2011 level 6; a programme of at least 3 years). ML engineering draws from a wider range of degree backgrounds than most ICT roles:
B.Tech / B.E. in Computer Science or IT (4 years) — the most common pathway. Qualifies directly if your university is H+ in anabin.
M.Tech or MSc in Computer Science, Data Science, or Artificial Intelligence — qualifies as a master's equivalent. MSc in AI from IITs and NITs is typically listed in anabin.
MSc or B.Sc. in Statistics or Applied Mathematics — qualifies if your institution is H+ and your degree is listed as "entspricht". Statistics degrees from IITs and central universities are well-recognised. A ZAB Zeugnisbewertung is recommended for Statistics degrees from smaller institutions.
MSc in Physics or applied sciences with significant ML coursework — borderline; outcome depends on programme content. A ZAB evaluation is recommended.
Recognition is verified through the anabin database. Your university must be rated H+; your degree must appear as "entspricht" or "gleichwertig". If your university is H+/-, the specific degree programme must be listed in the comments field.
Route 2: IT specialist without a degree (§ 18g(2))
If you have at least 3 years of ML/AI experience at university-graduate level within the last 7 years, you can qualify without a formal degree. The role must fall under ISCO-08 group 25 and your salary must meet the €45,934.20 threshold.
ML engineering roles typically qualify — designing and training models, building ML pipelines, deploying models to production, and maintaining ML infrastructure are all graduate-level work. Data labelling, basic analytics, or BI reporting roles that do not involve engineering-grade ML systems may not meet the "university-graduate level" standard.
Document checklist (India → Germany, 2026)
For Route 1 (with degree):
- Valid passport (issued within 10 years, at least 2 empty pages)
- Degree certificate (B.Tech, MSc, etc.)
- Mark sheets for every semester
- Confirmation from your university that you studied in regular (on-site) mode
- anabin printouts for your university and degree — or ZAB Statement of Comparability if your Statistics, Mathematics, or Physics programme is not listed
- Employment Declaration (Erklärung zum Beschäftigungsverhältnis) completed by your German employer
- Health insurance certificate from your German employer's insurer
For Route 2 (IT exception — no degree):
Replace degree documents with:
- Experience letters from each employer stating: job title, dates, specific technologies used (e.g. PyTorch, TensorFlow, scikit-learn, Hugging Face, MLflow, Kubeflow, Ray, Spark, Jupyter, pandas, NumPy, CUDA), description of ML responsibilities, and seniority level
- Salary slips or Form 16 for each employment period
- Optional but helpful: published papers, model cards, GitHub contributions, or design documents showing graduate-level ML scope
Note on apostille: Germany does not require or accept apostille on Indian documents. Do not pay for MEA apostille on your degree or experience letters.
After approval: settlement permit timeline
- 21 months of Blue Card employment + German at B1 → settlement permit
- 27 months of Blue Card employment + German at A1 → settlement permit
Your spouse has full work rights in Germany from day one.
Common mistakes
1. Non-standard title not described in the Employment Declaration. If your title is "Applied Scientist", "Research Engineer", or "AI Engineer", the Erklärung zum Beschäftigungsverhältnis must describe the technical work content so the Federal Employment Agency can confirm group 25 classification. Pure research roles in academia may be classified differently.
2. Statistics or Mathematics degree assumed to be automatically listed in anabin. Many Statistics and Mathematics programmes — especially from newer institutions — are not individually listed. Verify your specific programme in the anabin degree database before your visa appointment.
3. Route 2 experience letters that list frameworks without describing model scope. Letters for the IT exception must describe what you built: model types, dataset scale, production deployment, inference latency requirements, and your seniority. "Used PyTorch and TensorFlow" without context does not establish graduate-level work.
4. Using the 2025 salary threshold. The 2025 shortage threshold was €43,759.80. The binding 2026 figure is €45,934.20.
When you need a lawyer
Most applications do not require a lawyer. Consider one if:
- Your degree is in Statistics, Mathematics, or Physics and you are uncertain about the ZAB outcome
- You are applying via the IT exception and your role boundary is ambiguous — e.g. a data scientist role where the ML engineering component is unclear
- Your title is "Applied Scientist" or "Research Scientist" and you are concerned about ISCO-08 classification
- You are 45 or older — there is an additional pension provision requirement if your salary is below €55,770
We are not a law firm and this page does not constitute legal advice.
Frequently asked questions
Does an MSc in Data Science qualify for the EU Blue Card?
Yes, if it is recognised in Germany. An MSc in Data Science from an H+ university with the degree listed as "entspricht" in anabin qualifies under Route 1. If your programme is not listed — common for programmes launched after 2018 — a ZAB Statement of Comparability resolves this in approximately 2 weeks with a German work contract attached.
What ISCO-08 code does "ML Engineer" fall under?
Typically ISCO-08 unit 2519 (software and applications developers NEC) or unit 2521 (database designers and administrators), depending on whether your role is model-development or data-platform oriented. All are within group 25 — a shortage occupation — so the lower threshold and the IT exception both apply.
Can I apply as an ML engineer without a formal degree?
Yes, via the IT exception under § 18g(2). You need at least 3 years of ML/AI engineering experience at university-graduate level within the last 7 years, a salary at or above €45,934.20, and a role within ISCO-08 group 25. Strong experience documentation — describing models built, datasets used, and production deployment scope — is critical.
What technologies should I list in my experience letter?
Be specific. For ML roles: PyTorch, TensorFlow, scikit-learn, Hugging Face Transformers, MLflow, Kubeflow, Ray, Spark MLlib, Jupyter, pandas, NumPy, CUDA, Docker, Kubernetes. For LLM/GenAI roles: fine-tuning, RLHF, vLLM, LangChain, vector databases (Pinecone, Weaviate, pgvector). Generic terms like "ML frameworks" weaken the letter.
How long does the EU Blue Card take to process for Indian applicants?
The Federal Foreign Office publishes "up to 3 months, occasionally longer". The fast-track § 81a Vorabzustimmung procedure reduces total processing time to approximately 6 weeks and costs an additional €411.
Sources
- § 18g AufenthG — EU Blue Card — Bundesministerium der Justiz
- EU Blue Card — Make it in Germany — Federal Government
- anabin database — KMK / ZAB
- § 18c AufenthG — Settlement permit — post-March 2024 reform
We are not a law firm. This page provides general information only, not legal advice. German immigration law changes regularly — always verify current rules with the relevant German mission before applying.